Heterogeneous Response of Antimitochondrial Autoantibodies and Bile Duct Apical Staining Monoclonal Antibodies to Pyruvate Dehydrogenase Complex E2: The Molecule Versus the Mimic CHRIS MIGLIACCIO,1 JUDY VAN DE WATER,1 AFTAB A. ANSARI,2 MARSHALL M. KAPLAN,3 ROSS L. COPPEL,4 KIT S. LAM,1 RICHARD K. THOMPSON,5 FREDA STEVENSON,5 AND M. ERIC GERSHWIN1 The 2-oxo-acid dehydrogenase complexes and, in particular, the E2 component of the pyruvate dehydrogenase complex (PDC) are the target of antimitochondrial antibodies (AMA). More than 95% of primary biliary cirrhosis (PBC) patients have detectable levels of autoantibodies to PDC-E2 and in general these react with a region of the molecule that contains the prosthetic group lipoic acid (LA). LA is vital to the function of the enzyme, although there is conflicting evidence as to whether its presence is required for PDC-E2 recognition by AMA. Some, but not all, monoclonal antibodies (mAbs) to PDC-E2 produce an intense staining pattern at the apical surface of bile duct epithelial cells (BEC) in patients with PBC, and it has been argued that the molecule at the apical surface of PBC bile duct cells is a modified form of PDC-E2 or a cross-reactive molecule, acting as a molecular mimic. Herein, we characterize the epitopes recognized by 4 anti–PDC-E2 mAbs that give apical staining patterns (3 mouse and 1 human). In particular, by using a combination of recombinant antigens, competitive inhibition assays, and a unique peptide-on-bead assay, we determined that these apically staining mAbs recognize 3 or 4 distinct epitopes on PDC-E2. More importantly, this suggests that a portion spanning the entire inner lipoyl domain of PDC-E2 can be found at the BEC apical surface. In addition, competition assays with patient sera and a PDC-E2–specific mAb showed significant epitope overlap with only 1 of the 3 mouse mAbs and showed a differential response to the peptide bound to Abbreviations: PBC, primary biliary cirrhosis; AMA, antimitochondrial antibodies; PDC, pyruvate dehydrogenase complex; BEC, bile duct epithelial cells; mAbs, monoclonal antibodies; ILD, inner lipoyl domain; OLD, outer lipoyl domain; GST, glutathioneS-transferase; LA, lipoic acid; PSC, primary sclerosing cholangitis; HRP, horseradish peroxidase; ABTS, 2-azino-bis-thiosulfonate; ELISA, enzyme-linked immunosorbent assay; PBS, phosphate-buffered saline; OD, optimal dilution; Dde, N-1-(4,4-dimethyl-2,6dioxocyclohexylidene)ethyl; HOBt, N-hydroxybenzotriazole; DIC, diisopropylcarbodiimide; DMF, dimethyl formamide. From the 1Divisions of Rheumatology, Allergy & Clinical Immunology, and Hematology & Oncology, Department of Internal Medicine, University of California at Davis, Davis, CA; 2Department of Pathology, Emory University School of Medicine, Atlanta, GA; 3New England Medical Center Hospital, Boston, MA; 4Department of Microbiology, Monash University, Clayton Victoria, Australia; and 5Molecular Immunology Group, Tenovus Laboratory, Southamptom University Hospitals Trust, Southamptom, UK. Received November 13, 2000; accepted February 8, 2001. Supported in part by NIH DK39588. Address reprint requests to: Judy Van de Water, Ph.D., Division of Rheumatology, Allergy and Clinical Immunology, University of California at Davis School of Medicine, TB 192, One Shields Avenue, Davis, CA 95616. E-mail: javandewater@ucdavis.edu; fax: 530-752-4669. Copyright © 2001 by the American Association for the Study of Liver Diseases. 0270-9139/01/3304-0005$35.00/0 doi:10.1053/jhep.2001.23783 beads. These findings further highlight the heterogeneous response of patient autoantibodies to the inner lipoyl domain of PDC-E2. (HEPATOLOGY 2001;33:792-801.) Primary biliary cirrhosis (PBC) is a destructive autoimmune disease of intrahepatic bile ducts characterized by inflammation of the portal triads, fibrosis, and the presence of antimitochondrial antibodies (AMA).1,2 The major autoantigens recognized by AMA are related members of the 2-oxoacid dehydrogenase complexes including the pyruvate dehydrogenase complex (PDC-E2), the branched chain ketoacid dehydrogenase complex-E2, and the 2-oxo-glutarate dehydrogenase complex-E2. It has been hypothesized that a molecular mimic of the PDC-E2 molecule is expressed at high levels in the apical region of biliary epithelial cells (BEC) in PBC.3-5 We have previously described a series of 8 monoclonal antibodies (mAbs) against the mitochondrial antigens of PBC that produce this disease-specific staining pattern.6,7 Essentially, when probed with these disease-specific mAbs, small bile ducts in tissue from patients with PBC, but not controls, show intense staining at the apical surface of the cells lining the lumen. This pattern is in addition to the normal cytoplasmic pattern seen with mAbs against mitochondrial proteins. The fact that such apical staining is seen only with these select mAbs but not all mAbs that react with PDC-E2 has led to the hypothesis that a molecule(s) cross-reactive with PDC-E2 is located at the apical surface of PBC BEC.3,4 Extensive efforts have been made to elucidate the AMA epitopes and link them to pathology. The consensus is that although the immunodominant epitope is located within the inner lipoyl domain (ILD), a cross-reactive epitope that is less well recognized by AMA is localized to the outer lipoyl domain (OLD) of PDC-E2.8,9 In addition, cross-reactivity between anti–PDC-E2 antibodies and the E3-binding protein (previously referred to as protein X) has been shown.10-12 There is also strong evidence supporting the view that the BEC are the targets of immune-mediated destruction.13 In the current study, we analyzed the epitope specificity of 4 apically staining anti–PDC-E2 mAbs in comparison with a panel of other mAbs that include anti–PDC-E2 mAbs that do not differentially stain PBC tissues. Using synthetic peptides, recombinant antigens, and competitive binding immunoassays we were able to elucidate the nature of the epitopes recognized by these 4 mAbs. Despite the fact that each of these mAbs produces an identical apical pattern in PBC BEC, they do not recognize the same epitope. The likelihood that these mAbs are binding to different sites on the apical antigen is strong 792 HEPATOLOGY Vol. 33, No. 4, 2001 and suggests that a larger fragment of PDC-E2 is sequestered at the site of immune-mediated destruction in PBC livers. These findings question the presence of a molecular mimic of PDC-E2 localized to the apical surface of BEC and further substantiate the theory that disease-specific apical staining is due to PDC-E2, or a fragment thereof, that is present at the apical surface of BEC. MATERIALS AND METHODS Recombinant Proteins. The recombinant proteins used herein include the following: the lipoylated ILD corresponding to residues 128 to 229, ILD 128 to 229 without lipoic acid, and the OLD corresponding to residues 1 to 98 with lipoic acid.14,15 All the above antigens were expressed using the pGEX expression system previously described.16 Antibodies and Sera. Mouse mAbs were generated from BALB/c mice.6 The mAbs used for this study include 17 anti–PDC-E2 antibodies and, for the purpose of controls, 1A1, which reacts to glutathione S-transferase (GST). Rabbit anti–lipoic acid (LA) polyclonal antibody, coined CW38, was generated by immunization of New Zealand white rabbits with LA-KLH conjugate.17 Mouse mAbs were purified from either hybridoma supernatant or ascites. The human mAb, PD2, was generated via immortalization of a plasma cell from a PBC patient.12 Sera from 12 patients with PBC with previously defined reactivity against beef heart mitochondria and, for the purpose of controls, 4 patients with other autoimmune diseases (1 primary sclerosing cholangitis [PSC], 1 progressive systemic sclerosis, 2 systemic lupus erythematosus), and 4 healthy donors were used in this study. The mAb 2H4 was biotinylated using the EZ-Link Biotin-LCHydrazide kit (Pierce, Rockford, IL). Profile of Monoclonal Antibody Reactivity to PDC-E2. To determine reactivity against the PDC-E2 recombinant protein, purified recombinant PDC-E2 was used to coat 96-well microtiter plates. Several forms of the PDC-E2 fragmented antigens (described above) were used. The target antigens were each diluted to 1 mg/mL in 10 mL of carbonate coating buffer. Plates were coated with 100 mL of each of these antigens (0.1 mg/well) and stored at 4°C overnight. Coated plates were blocked with 3% milk at 200 mL/well for greater than 30 minutes at room temperature. After blocking, diluted primary antibody was added and incubated for 1 hour at room temperature. After washing, ABComplex (Vector Laboratories, Burlingame, CA) for biotinylated primary antibodies, or horseradish peroxidase (HRP)conjugated secondary antibodies (Zymed, South San Francisco, CA), were added and incubated at room temperature for 30 to 60 minutes. The plates were then washed and the substrate, pNPP (Vector Laboratories) in the case of biotinylated antibodies or 2-azino-bis-thiosulfonate (ABTS) for peroxidase-conjugated antibodies, was added and the plates read at 405 nm. Affinity Elution. To determine the role affinity plays in epitope recognition and apical staining, we used an enzyme-linked immunosorbent assay (ELISA) elution technique.18,19 This technique uses NH4SCN at serial concentrations to elute off antibodies in a standard ELISA. As described above for a standard ELISA, 96-well microtiter plates were coated with lipoylated ILD (0.1 mg/well). Plates were blocked with 3% milk in phosphate-buffered saline (PBS) before the addition of diluted primary antibody. Optimal dilutions of primary antibodies were determined in a previous ELISA. After an hour of incubation with primary antibody, the plates were washed (3 times) with PBS-Tween. The salt NH4SCN was then added in serial dilutions (8 mol/L, 2.7 mol/L, 0.9 mol/L, 0.3 mol/L, 0.1 mol/L) and incubated at room temperature for 15 minutes, then washed (3 times) with PBS-Tween. After washing, the secondary antibody (HRP-conjugated at 1:2,000) was added and incubated for 1 hour at room temperature. The plates were then washed, the substrate ABTS was added, and the plates were read at 405 nm. Affinities were determined by calculating the concentration of NH4SCN required to reduce the initial absorbance of an antibody (in the absence of salt) by 50%. MIGLIACCIO ET AL. 793 Competition Assays. For competition assays, before the addition of the test antibody (2H4, PD2, C355.1, 4C8, 1F1), blocking antibodies were added and incubated on the plates for 1 hour at room temperature. The blocking antibodies (2H4, 4C8, C355.1, PD2, 1F1, 1A1, 4H4, or patient sera) were serially diluted and included control wells containing no blocking antibodies. After incubation with the blocking antibodies, the plates were washed 3 times with PBS-Tween. The test antibody was then added to wells at an optimal dilution, and the protocol for a standard ELISA, as described above, was followed. Optimal dilutions of test antibodies were determined by screening serially diluted mAbs on a standard ELISA. The “optimal dilution” (OD) was defined as the highest possible dilution that still maintained a strong level of reactivity to the coated antigen. Percent inhibition was calculated by comparing the OD values of the blocked wells with the OD values of the unblocked wells. The calculation was performed using the following equation: ~OD U 2OD B !/OD U 3100 where ODU is the value for the unblocked reading, and ODB is the value for the blocked reading. Results are reported as normalized values as related to the experimental maximum inhibition of one mAb competing against itself. For our purposes, we competed purified 2H4 with biotinylated 2H4. Over the course of several assays, 2H4 consistently produced a maximum inhibition of 59%. Therefore, for normalization of our results, inhibition of 59% is considered as 100% inhibition, or epitope overlap. Peptide Synthesis. Standard solid-phase peptide synthesis was performed on TentaGel Resin S Amino-NH2 (Rapp Polymere, Tubingen, Germany). A list of the peptides and controls used for mapping is described in Table 1. All Fmoc amino acids, with standard side chain protecting groups (with the exception of lysine), were obtained from Advanced ChemTech (Louisville, KY), Bachem (Torrance, CA), or Propeptide (Vert-le-Petit, France). Lysine with the side-chain protecting group of N-1-(4,4-dimethyl-2,6-dioxocyclohexylidene)ethyl (Dde) was required for selective deprotection and addition of the prosthetic groups. N-hydroxybenzotriazole (HOBt), piperidine, and diisopropylcarbodiimide (DIC) were obtained from Advanced ChemTech. Peptide-beads were synthesized according to Lam et al.20 with modification. HOBt and DIC were used as activating reagents and 4-fold excess of amino acids were used for each coupling step. Each TABLE 1. PDC-E2–Derived and Control Peptides Synthesized on TentaGel Beads 1 2 3 4 5 A A A A A A A A A A E E E E E E E E E E E I I I I I I I I I I I I E E E E E E E E E E E E E T I T T T T T T T T T T T T 6 7 8 9 10 11 12 13 14 15 16 17 D D D A D D D D D D D D D D D A D F K K K K R K K K K K K K K K K K E K A A A A A D A A A A A A A A A A A G T T T T T T I T T T I I I I I I N I I G G G G G G A G F F F F F F D E V Q E E T T T T T T L I I I I I I V G G G G G G L F F F F F F I E E A V V F Q Q S E K Q E E Y NOTE. The boldfaced lysine (K) is the site of LA attachment. The italicized residues are substitutions in the native sequence. The last peptide on the chart is a control sequence derived from bovine serum albumin. 794 MIGLIACCIO ET AL. HEPATOLOGY April 2001 Fmoc amino acid was first mixed with HOBt in dimethyl formamide (DMF) and added to the TentaGel S NH2 resin (0.1-0.5 g). DIC was then added immediately to the beads. The beads were mixed for more than 1 hour at room temperature. Twenty-five percent piperidine in DMF (vol/vol) was used to remove Fmoc groups according to the following method: incubation of the beads for 5 minutes on a rocker, followed by the removal of the fluid and the addition of fresh piperidine and mixing for an additional 15 minutes. Free amino groups were determined by the Kaiser test.21,22 After the coupling of the terminal amino acid to the peptide, the beads were washed with DMF, and terminal Fmoc groups were removed with piperidine as described above. A small aliquot of each peptide bead was then set aside for microsequencing. The N-a amino groups were acylated with 15% acetic acid with HOBt and DIC in dichloromethane for 15 minutes. The side-chain protecting group Dde on the lysine residue was specifically removed by treatment with 2% hydrazine in DMF for 3 minutes (2 times). The coupling of LA (6,8-thioctic acid oxidized) (Sigma, St. Louis, MO) was performed according to a protocol similar to the amino acid coupling procedure described above (overnight) with the addition of HBtU and N-ethylmaleimide. After LA coupling to the lysine, all remaining side-chain protecting groups were removed with a 95% trifluoroacetic acid treatment. Beads were washed with DMF (3 times), methanol (3 times), and dichloromethane (6 times), and used in our study. For controls, we used 3 different peptides. Two were derivatives of the native sequence (DKATIGFEVQEE). One of the derivative peptides incorporated a single residue substitution of an alanine (A) for the aspartic acid (D) adjacent to the lysine (K) (AKATIGFEVQEE). The other natively derived peptide contained a switch of the lysine (K) with the glutamic acid (E) at the C-terminal end of the peptide (DEATIGFEVQKE). The third control peptide used a 12mer from serum albumin, which contained a lysine (K) at the same position as our native peptide (FKGLVLIAFSQY). Each of these control peptides were either coupled with LA or left with a free amine group. In addition, as a control, LA was coupled directly to the beads, without any amino acids. Bead ELISA. Bead-bound peptides were used to determine binding of the appropriate antibody to the various peptides, via an ELISAbased assay, performed in mini-columns (Wallace Inc., Akron, OH). Beads (500-1,500 per column in a fixed volume of bead suspension) were blocked for 30 minutes at room temperature with 1% bovine serum albumin in PBS. After blocking, 500 mL of the primary antibody dilution was added per column and incubated at room temperature for 1 to 3 hours. HRP-conjugated secondary antibody (Zymed; or Biosource International, Fullerton, CA) was used at a predetermined optimal dilution of 1:3,000 and incubated with beads for 1 hour at room temperature. Washing was performed after the addition of the primary and secondary antibodies using 0.5%Tween in PBS (4-5 times per column). After the washing of the secondary antibody, 200 to 300 mL of PBS was added to each column and the beads were resuspended and transferred to a 96-well microtiter plate in duplicate or triplicate (100 mL per well). A total of 100 mL of substrate, ABTS (Sigma), was added to each well and allowed to incubate at room temperature for 5 to 15 minutes, and optical density readings were performed on a microtiter plate reader at 405 nm. Structural Analysis. The sequence for human PDC-E2 was obtained from the NCBI database. The amino acid sequence was then submitted to the Swiss-Model program for structure prediction. The structure generated by Howard et al. (PDB# 1FYC) was downloaded for analysis of the comparative epitopes of the antibodies.23 RESULTS Profile of mAb Reactivity to PDC-E2. The pattern of reactivity of the 17 PDC-E2–specific mAbs was analyzed and appeared TABLE 2. Categories of Reactivity of Monoclonal Antibodies to PDC-E2: Inner Versus Outer Lipoyl Domain Group* Antibody Isotype ILD (128-229) OLD (1-91) Apical† I 3G11 3H5 3F2 2D5 6H1 IgMk IgMk IgMk IgG2bk IgG2bk 0.562 6 0.141 0.285 6 0.042 0.217 6 0.020 0.115 6 0.055 0.103 6 0.032 0.064 6 0.005 0.186 6 0.057 0.071 6 0.001 0.060 6 0.005 0.065 6 0.001 2 2 2 2 2 II 6C10 IgG2bk 0.093 6 0.013 0.067 6 0.003 2 III C150 IgG1 0.066 6 0.003 1.627 6 0.417 2 IV C355.1 1A3 6C5 3E4 2B7 IgG2b IgG1k IgG2bk IgG1k IgG1k 1.194 6 0.027 0.469 6 0.030 0.809 6 0.167 0.201 6 0.004 0.184 6 0.037 0.064 6 0.009 0.070 6 0.007 0.069 6 0.001 0.068 6 0.005 0.064 6 0.001 1 2 2 2 2 V 4C8 1F1 4H2 IgG2bk IgMk IgMk 1.111 6 0.142 1.064 6 0.214 0.570 6 0.040 0.112 6 0.013 0.132 6 0.022 0.487 6 0.123 1 2 2 VI 2H4 PD2 IgG1k IgG2 1.635 6 0.101 1.731 6 0.032 1.670 6 0.217 0.454 6 0.137 1 1 Controls PBC patients‡ CW38 (lipoic acid) 1A1 (GST) poly poly IgG2bk 1.714 6 0.018 1.977 6 0.156 0.064 6 0.001 1.877 6 0.120 2.148 6 0.011 0.177 6 0.031§ NA NA 2 NOTE. All values are average of 4 assays (OD) 6 SEM. *Grouped by reactivity profiles. The members of each group are listed in descending order of absorbance. †Refers to disease-specific apical pattern. ‡PBC patients’ data in this table represent an average of patient nos. 3-12. §The OLD antigen is expressed in a GST-fusion system, thus this value expresses a small amount of GST contamination. HEPATOLOGY Vol. 33, No. 4, 2001 MIGLIACCIO ET AL. 795 TABLE 3. Categories of Reactivity of Monoclonal Antibodies to PDC-E2: Lipoic Acid Requirement Group* Antibody Isotype ILD, LA1 (128-229) ILD, LA2 (128-229) Apical† I 3G11 3H5 3F2 2D5 6H1 IgMk IgMk IgMk IgG2bk IgG2bk 11 11 1 1/2 1/2 2 2 2 2 2 2 2 2 2 2 II 6C10 IgG2bk 2 111 2 III C150 IgG1 2 2 2 IV C355.1† 1A3 6C5 3E4 2B7 IgG2b IgG1k IgG2bk IgG1k IgG1k 111 11 111 1 1 111 11 1/2 1 1/2 1 2 2 2 2 V 4C8† 1F1 4H2 IgG2bk IgMk IgMk 111 111 11 111 111 1/2 1 2 2 VI 2H4† PD2† IgG1k IgG2 111 111 2 2 1 1 Controls PBC patients‡ CW38 (lipoic acid) 1A1 (GST) poly poly IgG2bk 111 111 2 111 2 2 NA NA 2 *Grouped by reactivity profiles. The members of each group are listed in descending order of absorbance. †Disease-specific apical staining pattern on tissue (bile ducts). ‡PBC patients’ data in this table represent an average of patients 3-12. to show 6 distinct patterns. Patterns were identified based on reactivity to ILD versus OLD (Table 2), and ILD with or without LA (Table 3). Similarities and differences between the ILD and OLD are illustrated in Fig. 1. Group I consists of 5 mAbs that react only to the lipoylated ILD and range from strong (OD3G11 5 0.562) to weak (OD6H1 5 0.103). Groups II and III consist of a single mAb each, with 6C10 (group II) having strong reactivity to the unlipoylated ILD only and C150 (group III) having strong reactivity to the OLD. There are no disease-specific apically staining mAbs in these first 3 groups. Each of the remaining 3 groups (IV-VI) contains at least 1 apically staining mAb. Group IV consists of 5 mAbs that react only with the ILD, regardless of the presence of LA. The disease-specific member of this group is C355.1. The reactivity in group IV varies greatly with values ranging from 1.194 FIG. 2. Relative affinities of the 5 anti–PDC-E2 mAbs. The antibodies used for this experiment were the 4 disease-specific mAbs (4C8 [solid triangle, dotted line], 2H4 [solid square, solid line], PD2 [asterisk, solid line], and C355.1 [solid diamond, dashed line]) and 1 control (1F1 [“3,” solid line]). Using a standard ELISA coated with recombinant PDC-E2, the mAbs were eluted off of the antigen with serial dilutions of the salt (NH4)SCN. Note the relatively low affinity of PD2, whereas 4C8, 1F1, C355.1, and 2H4 all have comparable affinities with only subtle differences. Values depicted represent the mean of 3 determinations 6 SEM. (ODC355.1) to 0.184 (OD2B7) for the lipoylated and 1.265 (ODC355.1) to 0.134 (OD2B7) for the unlipoylated ILD. Group V consists of 3 mAbs that react with both OLD and ILD (with or without LA). All 3 react significantly with the lipoylated ILD (OD range of 1.111-0.570). However, reactivity of group V mAbs to the OLD and unlipoylated ILD varies, with OLD reactivity ranging from 0.112 (OD4C8) to 0.487 (OD4H2) and unlipoylated ILD reactivity ranging from 0.114 (OD4H2) to 0.739 (OD1F1). Lastly, group VI consists of 2 mAbs, one mouse and one human, that react only with the lipoylated ILD and OLD. Both exhibit strong reactivity to the lipoylated ILD (OD . 1.6), but vary in OLD reactivity with an OD range of 1.670 (OD2H4) to 0.454 (ODPD2). Affinity Elutions. The results of the affinity determinations are depicted in Fig. 2. Using serially diluted NH4SCN to test for antibody affinity, we were able to determine the relative affinities for the apically staining mAbs. The human mAb PD2 eluted at a lower concentration of NH4SCN compared with the other 3 mouse mAbs. These data indicate that, although the initial absorbance of the human mAb was reduced below 50% at a salt concentration of 0.3 mol/L, the initial absorbance of the mouse mAbs were not reduced below 50% until 2.7 mol/L with only subtle differences between them. FIG. 1. Homology analysis of the ILD and OLD. The top sequence corresponds to the OLD of PDC-E2 residues 1-98; the lower sequence corresponds to the ILD of PDC-E2 residues 128-225. LA indicates the lipoic acid– binding residue. The circled residues are the 13 nonconserved differences between the ILD and the OLD. An example of a nonconserved difference is a hydrophobic versus a hydrophilic residue, whereas a conserved difference would be a valine versus an isoleucine. 796 MIGLIACCIO ET AL. HEPATOLOGY April 2001 TABLE 4. Profile of PBC and Control Patient Serum Reactivity Blocking Sera PBC-1 PBC-2 PBC-3 PBC-4 PBC-5 PBC-6 PBC-7 PBC-8 PBC-9 PBC-10 PBC-11 PBC-12 Diseased controls PSC Normal control PDC-E2 Titer 1:103 Antigen Specificity* 1:103 .1:108 .1:108 .1:108 .1:108 .1:108 .1:108 .1:108 .1:108 .1:108 .1:108 Weak PDC-E2 PDC-E2 PDC-E2, E3BP, BCOADC-E2 PDC-E2, E3BP PDC-E2, E3BP, BCOADC-E2 PDC-E2, E3BP, BCOADC-E2 PDC-E2, E3BP PDC-E2, E3BP, BCOADC-E2 PDC-E2, E3BP, BCOADC-E2 PDC-E2, E3BP PDC-E2, E3BP PDC-E2, E3BP — — — — *Determined by probing immunoblot of beef heart mitochondrial preparations separated on 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis. Competitive Binding Analysis of the 4 Disease-Specific mAbs. Since all 4 mAbs appeared to show essentially the same disease-specific apical staining of BEC, it was reasoned that these antibodies could potentially bind the same epitope. Thus, antibody competition assays were performed to address this issue. The mAb 2H4 was biotinylated and used as the test antibody as it is the only mAb in this study with a defined epitope. As expected the unconjugated 2H4 showed competitive binding with itself. However, none of the other blocking mAbs resulted in a significant decrease of 2H4 signal. When a similar competition assay was performed using the human mAb PD2 as the test antibody a different pattern emerged. Three antibodies (1F1, 2H4, and C355.1) showed minimal blocking that quickly diminished, whereas the fourth mAb, 4C8, produced a high level of competition with PD2 that was sustained through the highest dilution (.1:3,000). Reverse inhibitions were not performed. PDC-E2 binding for mAbs reactive against PDC-E2 was likewise compared with patient sera. Profiles of patient sera used for competition assays are depicted in Table 4. Competition of patient sera with 4C8 resulted in strong inhibition of 4C8 binding to PDC-E2. While sera from 2 patients with PBC, the non–liver disease patients, and the healthy controls all resulted in no detectable blocking of 4C8 at any dilution, 10 of 12 PBC sera generated complete blocking of 4C8 binding to PDC-E2. These 10 patients blocked 4C8 binding more than 100% at dilutions of 1:50 and 1:100 with an average slope of 7.15 (Table 5 and Fig. 3). The mAb 1F1, a nonapically staining anti–PDC-E2 mAb, was also analyzed for competitive inhibition with patient sera. The same 10 patients with PBC that were found to block 4C8 likewise efficiently blocked 1F1. Four of these 10 samples effectively blocked 1F1 at a greater than 100% efficiency at the lower dilutions, maintaining inhibition above 48% up to dilution 1:3,200 (Table 4 and Fig. 3). The remaining 6 patient samples blocked efficiently (97% to greater than 100%) at dilutions of 1:50 and 1:100, but soon dropped below 10% at 1:800 or 1:1,600. In addition, 1F1 also experienced a low level (33%) of blocking from one of the PSC sera at 1:50 but was lost at a 1:100 dilution. Competition of patient sera with C355.1 and 2H4 resulted in minimal block- ing with nothing of significance at dilutions beyond 1:20 and 1:50, respectively (data not shown). The reverse of these inhibition experiments was also performed where the mAbs were used to block patient sera binding. None of the mAbs tested, including 4C8, 2H4, C355.1, and 1F1, resulted in a detectable decrease in patient serum binding. This was expected because PBC sera contain antibodies with multiple antigen and epitope specificities. Peptides. Using peptides synthesized on beads, a potentially complete epitope was determined for 2H4. This mAb reacted with peptides containing the residues aspartic acid (D) and lysine (K), when they were coupled with the prosthetic group lipoic acid (LA). Deletions and substitutions of amino acids within the 12mer supported this conclusion. The binding of 2H4 was not significantly altered after systematic deletion of a single residue per peptide from either the C-terminus or the N-terminus. Systematic substitutions of the 4 residues within the LA-binding region (TDKA) resulted in a significant decrease in 2H4 binding when either the lysine (K) or the aspartic acid (D) were targeted, but not threonine (T) or alanine (A). However, substitutions of the residues of 1 b-strand within the loop (to disrupt conformation) did not result in any significant change in the binding of 2H4 (Table 6). None of the control peptides resulted in detectable reactivity with 2H4. With the exception of the rabbit anti-LA antibody (CW38), none of the other mAbs used reacted to peptides with any level of consistency. CW38 consistently recognized only the peptides and beads coupled with LA. DISCUSSION Although the presence of high-titer AMA is the hallmark of PBC, the relationship between autoantibody and pathology is not fully understood. Previous studies on identification of the TABLE 5. Inhibition of PDC-E2–Specific mAbs 4C8 and 1F1 With Serum From Patients With PBC and Controls 4C8 Inhibition Blocking Sera PBC-1 PBC-2 PBC-3 PBC-4 PBC-5 PBC-6 PBC-7 PBC-8 PBC-9 PBC-10 PBC-11 PBC-12 Diseased controls PSC PSS & SLE Normal controls 1F1 Inhibition Maximum % (Dilution) Slope Maximum % (Dilution) Slope 0 0 90 (1:50)* 74 (1:50)* 74 (1:50) 79 (1:100)* 81 (1:100) 82 (1:100)* 78 (1:50)* 81 (1:50)* 80 (1:50) 84 (1:100)* NA NA 28.78 26.39 27.47 26.85 210.87 24.39 24.51 23.70 210.82 27.67 0 0 70 (1:200) 70 (1:50)* 72 (1:50)* 64 (1:200)* 58 (1:50)* 74 (1:50) 59 (1:50) 69 (1:50) 63 (1:50) 80 (1:50) NA NA 22.36 24.79 27.15 22.14 210.07 213.54 29.83 211.44 210.15 214.74 0 0 0 NA NA NA 22 (1:50) 0 0 21.81 NA NA NOTE. The efficiency of the assay was determined by blocking 2H4-biotin with nonbiotinylated 2H4. The result was a decrease in OD values by 59%. Using this as a reference, any values .59% blocking could be considered to have completely overlapping epitopes. Abbreviations: PSS, progressive systemic sclerosis; SLE, systemic lupus erythematosus; NA, not applicable. *The antibody still blocked at the highest dilution (1:3200). HEPATOLOGY Vol. 33, No. 4, 2001 MIGLIACCIO ET AL. 797 FIG 3. Competition assays between sera from patients with PBC and mAbs. Using a standard ELISA protocol, the mAb binding was inhibited to varying degrees by serially diluted patient sera. Competition analysis with 10 patients are depicted on the graphs versus (A) 1F1 and (B) 4C8, where each patient is a separate curve. Note that blocking of 4C8 is relatively consistent among patient samples; whereas, the same 10 patients blocking 1F1 resemble 2 distinct profiles: 4 that block in a manner similar to those against 4C8, and 6 that block at a reduced level. Values depicted represent the mean of 3 determinations 6 SEM. T cell and AMA epitopes have focused on the major domains of the PDC-E2 as well as the requirement of LA for epitope recognition.24-32 The consensus of these studies is that the immunodominant epitope is located within the ILD. We report herein on the characterization of 17 mAbs against the major autoantigen of PBC, PDC-E2, with particular emphasis on fine mapping of the epitopes that are recognized.6,12,15 Four of the 17 (C355.1, 2H4, 4C8, and PD2) have been shown to be disease specific as evidenced by their ability to stain PBC BEC but not control tissue in an apical pattern. The identity of the antigen at the apical surface has still not been established.3,4,33 However, we have previously reported that the apical staining is also seen with mAbs against the branched chain ketoacid dehydrogenase complex-E2 and the 2-oxo- glutarate dehydrogenase complex-E2 antigens.6 These mAbs do not cross-react with PDC-E2. In fact, of the 20 mAbs reactive to the other 2 mitochondrial enzymes associated with PBC, only 5 produce the disease-specific pattern. Thus, the material reactive in the BEC of PBC tissue expresses only some of the immunologic features of these 3 antigens. The intent in the current study was to further characterize the disease-specific PDC-E2 mAbs to determine the epitopes recognized by these antibodies and the relationship these epitopes share with each other and with epitopes recognized by patient sera AMA. Analysis of binding of PDC-E2 fragments by PDC-E2–specific antibodies led to the definition of 6 groups of mAbs. Of note is that there is no clear partition of apically staining mAbs into a single group; such reagents are TABLE 6. Mapping of mAb 2H4 Reactivity to the Inner Lipoyl Domain After Amino Acid Substitutions 2H4 Controls CW38 Rabbit Sera AEIEIDKATIGF AEIETAKATIGF AEIETDRATIGF AEIETDKDTIGF 0.895 6 0.138* 0.134 6 0.010† 0.115 6 0.003† 0.785 6 0.065 0.788 6 0.025 0.093 6 0.005 0.870 6 0.023 0.097 6 0.008 0.084 6 0.003 0.100 6 0.001 0.797 6 0.029 0.086 6 0.006 AEIETDKAINAD Native LA1 Native LA2 0.790 6 0.030 0.719 6 0.053 0.141 6 0.014 0.849 6 0.033 0.083 6 0.002 0.788 6 0.035 0.098 6 0.004 0.086 6 0.004 0.075 6 0.003 *Values given are the average of 3 assays 6 SEM. †Reactivities to these substitutions are significant (P , .01) when compared with the same antibody on native LA1 peptides. The substituted amino acid(s) for each peptide is underlined with the new residue inserted in the sequence. 798 MIGLIACCIO ET AL. HEPATOLOGY April 2001 FIG 4. Sequence of the OLD and ILD with putative epitope regions. The sequences depicted by boxes A and A9 form a region on the surface of the PDC-E2 molecule that is near the LA-binding site. The sequences depicted by box B form a region on the surface of PDC-E2 that is on the opposite side from A/A9. The sequences depicted by boxes C and C9 form a region on the surface of PDC-E2 next to region B and at the opposite end of the molecule from the LA-binding site. In addition, the C/C9 region contains the 3 amino acid differences between rat and human PDC-E2. According to the molecular model of the ILD, Tyr129, Val156, Val180, Gln181, Glu211, Ala212, Ala218, and Val225 are exposed on the surface of the enzyme. Of these, Tyr129, Val156, Glu211, and Ala212 are located at or near the residue differences between the rat and human proteins. found in groups IV, V, and VI. Comparing binding patterns of these apical-staining mAbs suggests that there are at least 3 or 4 different epitopes recognized. Three of the apically staining mAbs (4C8, 2H4, and PD2) cross-react with the ILD and OLD, whereas the fourth apical mAb (C355.1) does not. The epitope of C355.1 (group IV) is most likely distant to the LA-binding site because it does not require LA and does not inhibit 2H4 binding. Additionally, in our hands C355.1 performs markedly better in an ELISA and on fixed tissue than immunoblots, suggesting a conformational, rather than linear, epitope. Our putative epitope for C355.1 is located downstream of the LA-binding site in a region containing the least homology between the ILD and the OLD and corresponds with the C/C9 regions in Figs. 4 and 5. This region also contains the 3 residues that differ between the human and rat forms of the antigen. In addition, C355.1 appears to have no significant overlap with the epitope recognized by serum AMA, thus showing further evidence of an epitope distant from the LA-binding site. Group V mAbs react to both the ILD and OLD with 2 members, 1F1 and 4C8, binding strongly to the ILD regardless of the presence of LA. The direct competition and affinity experiments with 1F1 and 4C8 suggest a fair degree of epitope overlap for these 2 mAbs. In addition, competition between 4C8 and PD2 for PDC-E2 binding suggests that the epitope for 4C8 is more similar to the human mAb than 1F1. These results are reinforced by the ability of PBC sera to compete more effectively with 4C8 than 1F1. Theoretically, it is possible that the 1F1 epitope differs enough from the 4C8 epitope such that it is covered or hidden at the apical surface. This is supported by the fact that 1F1 does not apically stain the BEC in PBC patient tissue. The putative epitope for these mAbs is in the A/A9 region of our model (Figs. 4 and 5). This takes into account a distance from LA that allows the antibodies to remain unaffected by the prosthetic group, whereas PD2 can significantly overlap the epitopes of these mAbs and remain sensitive to LA. Placing the epitope in the A/A9 region allows for a minimal level of difference between the ILD and OLD, which could explain the weaker reactivity of the group V mAbs for the OLD. The epitopes for group VI would appear to be the closest in proximity to LA. The ability of these mAbs to recognize ILD and OLD suggests a conserved epitope. The epitope for 2H4 appears localized to the immediate LA-binding region, consisting of the residues DK-LA; however, data on the PD2 epitope are not as clear. Although PD2 requires the presence of LA for binding, it does not significantly bind to any of our peptides. This may suggest a larger and perhaps more conformationally complex epitope for the human mAb or may be caused by its relatively low affinity. Interestingly, 2H4 binding is not significantly inhibited by patient sera, suggesting that the majority of AMA do not bind directly to LA. In fact, PD2 may only represent a small subset of patient AMA that requires the presence of LA for recognition. Several lines of evidence suggest that AMA are composed of a heterogeneous group of antibodies with differing specificities, even though, overall, the AMA map to a relatively small region of PDC-E2. First, studies to determine the requirement of LA for AMA binding have long been controversial. Quinn et al. report the necessity of LA for serum AMA binding, whereas our own earlier result suggests otherwise.26,34 In addition, Thomson et al. generated human mAbs from patients with PBC and showed a requirement of LA for PD2, the reactivity of which has been confirmed by our data. However, it is important to note that pooled PBC sera binding was relatively unaffected by delipoylation.12 This again suggests a subset of AMA that are reactive to the ILD, only some of which require the presence of LA. We have also shown both here and in a recent study that delipoylation of PDC-E2 resulted in no significant reduction in binding of patient sera.35 Likewise, our attempts to block the binding of sera to ILD (LA1)-coated plates with purified, disease-specific mAbs proved fruitless, despite high dilutions of sera. Third, work done with patient sera and peptides has proven difficult. Previously, Van de Water et al. were able to effectively absorb reactivity using peptides, but only at high serum dilutions.36 In addition, unpublished data by our group have shown AMA reactivity to the 12mer peptide of PDC-E2 used herein to be dependent on the presence of LA but weaker than would be expected with sera at low dilutions. This suggests that the 12mer reactivity is present in a subset of patient AMA. Lastly, our most effective inhibition of mAbs by AMA was observed for 4C8 and 1F1, neither of which requires LA for binding. These observations suggest that patient reactivity to the ILD is heterogeneous and not exclusive to the LA-binding site. The appearance of heterogeneous reactivity has been proposed for another autoantigen of PBC, gp210,37 as well as described in other autoimmune diseases such as Grave’s disease, Goodpasture syndrome, and systemic lupus erythematosus.38-42 As noted earlier, of the 12 patients with PBC who HEPATOLOGY Vol. 33, No. 4, 2001 FIG. 5. Three-dimensional modeling of the linear sequence (Fig. 4) of the ILD is depicted. The linear sequence of the ILD was analyzed and epitopes corresponding to nonlinear sequences were determined. The A/A9 region is located structurally at the end of our 12mer peptide (-FEVQEE). Linearly, the 2 parts of the region are located on either side of the LA-binding site. Region B consists of a long linear sequence that folds back on itself to form a large portion of the surface of the ILD directly opposite of the A/A9 region. Region C/C9 is located structurally at the opposite end of the ILD as the LA-binding site. Linearly, region C/C9 consists of the residues on either side of region B. Region C/C9 includes the highest degree of difference between the ILD and the OLD, as well as the 3 residue differences between the human and rat antigens. (A) “Top-side” of the molecule herein; (B) “underside” of the molecule. were used for inhibition assays, 3 distinct levels of reactivity were apparent. Although 4C8 competes consistently and effectively with the same patient sera, as well as with the human mAb PD2, 1F1 is truly inhibitory with only 4 of the 12 patients. It is reasonable to consider that the epitope for 1F1 is such that it overlaps with multiple patient epitopes, including a putative epitope of the IgA isotype. If the antigen is complexed at the BEC surface with the AMA of the secretory MIGLIACCIO ET AL. 799 isotype, this would also explain the inability of 1F1 to stain apically. The presence of these multiple sets of AMA suggests a “localized heterogeneity” of epitopes. Although the AMA have been mapped to a narrow region of the ILD, there appears to be a variety of binding sites within this domain. The data presented in this study suggest that a portion spanning the ILD of PDC-E2 is present at the apical surface of BEC, although it is still not clear if the entire molecule is present. This is supported by the observation that the 4 apically staining mAbs appear to have epitopes that do not overlap significantly with one another and span most of the ILD, the portion of the molecule represented at the surface would appear to include the entire ILD. As to why the majority of antibodies to PDC-E2 do not generate an apical pattern, one possible explanation is that the epitopes recognized by the 4 disease-specific mAbs are the few sites on the molecule available for binding. The fact that our lone OLD-specific mAb (C150) does not apically stain and that C355.1 does not crossreact with the OLD, suggests that either the OLD is not present at the BEC surface or that the epitope for C150 is hidden or obstructed within a complex or altered conformation. However, because the other 3 mAbs cross-react with the OLD, it is possible that their epitopes in both domains are exposed at the apical surface or that the apical reactivity seen is caused by ILD binding alone. It may also be that some sites available for mAb binding are blocked by patient antibodies, leading to lack of apical staining by some mAbs. If this theory is true, it would lead to an abandonment of the “molecular mimic” theory at least in terms of a mimic acting as the target on BEC. However, a mimic may still be involved in breaking of tolerance, which then stimulates a response to PDC-E2 and the other mitochondrial antigens. Because these antigens are not seen at the surface of BEC in control liver, their appearance in PBC appears to be related to the disease process. How might this apical staining be related to the pathogenic process? In this regard we note the presence of immune complexes in the bile duct lumen. The dimeric form of the IgA isotype is transcytosed from the basal to the apical side of bile duct cells,43 and it is possible that IgA could bind to the antigen extracellularly before transcytosis.44,45 The concept of intracellular binding of antigens has been previously described for an epithelial virus during transcytosis.46 Moreover, in vitro work in our laboratory has described the colocalization of PBC serum IgA at the mitochondrial surface.47 The generally observed role of IgA in mucosal immunity is to bind antigens in the intestinal lumen and facilitate their clearance.48 In addition, it has been suggested that the generation of mucosally derived antibodies is separate from serum-derived antibodies.49,50 In a study of anti-sperm autoantibodies in infertile men, researchers found the reactivities of the IgA and IgG isotypes to be distinct from one another within each patient.49 It is possible that the IgA class of a patient AMA profile recognizes a different PDC-E2 epitope than IgG and IgM AMA. In addition, IgA is considered to be a poor inducer of inflammation to the point that it is sometimes referred to as “noninflammatory.”50-52 These properties would suggest that IgA complexed at the apical surface with PDC-E2 is not a direct inducer of inflammation in PBC. However, the presence of IgG AMA, as well as functional complement, in bile could have a major effect on the inflammatory process.7,53 The presence of disparate IgA and IgG epitopes on the same molecule 800 MIGLIACCIO ET AL. HEPATOLOGY April 2001 would allow IgG to bind to IgA-complexed antigen and thus more efficiently activate an inflammatory reaction. REFERENCES 1. Gershwin ME, Ansari AA, Mackay IR, Nishio A, Rowley MJ, Coppel RL. Primary biliary cirrhosis: an orchestrated immune response against epithelial cells. Immunol Rev 2000;174:210-225. 2. Kaplan MM. Primary biliary cirrhosis. N Engl J Med 1996;335:15701580. 3. Joplin R, Gershwin ME. Ductular expression of autoantigens in primary biliary cirrhosis. Semin Liver Dis 1997;17:97-103. 4. Joplin RE, Wallace LL, Lindsay JG, Palmer JM, Yeaman SJ, Neuberger JM. The human biliary epithelial cell plasma membrane antigen in primary biliary cirrhosis: pyruvate dehydrogenase X? Gastroenterology 1997; 113:1727-1733. 5. Van de Water J, Ansari AA, Surh CD, Coppel R, Roche T, Bonkovsky H, Kaplan M, et al. Evidence for the targeting by 2-oxo-dehydrogenase enzymes in the T cell response of primary biliary cirrhosis. J Immunol 1991;146:89-94. 6. Migliaccio C, Nishio A, Van de Water J, Ansari AA, Leung PS, Nakanuma Y, Coppel RL, et al. Monoclonal antibodies to mitochondrial E2 components define autoepitopes in primary biliary cirrhosis. J Immunol 1998; 161:5157-5163. 7. Van de Water J, Turchany J, Leung PS, Lake J, Munoz S, Surh CD, Coppel R, et al. Molecular mimicry in primary biliary cirrhosis. Evidence for biliary epithelial expression of a molecule cross-reactive with pyruvate dehydrogenase complex-E2. J Clin Invest 1993;91:2653-2664. 8. Koike K, Ishibashi H, Koike M. Immunoreactivity of porcine heart dihydrolipoamide acetyl- and succinyl-transferases (PDC-E2, OGDC-E2) with primary biliary cirrhosis sera: characterization of the autoantigenic region and effects of enzymatic delipoylation and relipoylation. HEPATOLOGY 1998;27:1467-1474. 9. Surh CD, Coppel R, Gershwin ME. Structural requirement for autoreactivity on human pyruvate dehydrogenase-E2, the major autoantigen of primary biliary cirrhosis. Implication for a conformational autoepitope. J Immunol 1990;144:3367-3374. 10. Palmer JM, Jones DE, Quinn J, McHugh A, Yeaman SJ. Characterization of the autoantibody responses to recombinant E3 binding protein (protein X) of pyruvate dehydrogenase in primary biliary cirrhosis. HEPATOLOGY 1999;30:21-26. 11. Surh CD, Roche TE, Danner DJ, Ansari A, Coppel RL, Prindiville T, Dickson ER, et al. Antimitochondrial autoantibodies in primary biliary cirrhosis recognize cross-reactive epitope(s) on protein X and dihydrolipoamide acetyltransferase of pyruvate dehydrogenase complex. HEPATOLOGY 1989;10:127-133. 12. Thomson RK, Davis Z, Palmer JM, Arthur MJ, Yeaman SJ, Chapman CJ, Spellerberg MB, et al. Immunogenetic analysis of a panel of monoclonal IgG and IgM anti-PDC-E2/X antibodies derived from patients with primary biliary cirrhosis. J Hepatol 1998;28:582-594. 13. Joplin RE, Neuberger JM. Immunopathology of primary biliary cirrhosis. Eur J Gastroenterol Hepatol 1999;11:587-593. 14. Liu S, Baker JC, Roche TE. Binding of the pyruvate dehydrogenase kinase to recombinant constructs containing the inner lipoyl domain of the dihydrolipoyl acetyltransferase component. J Biol Chem 1995;270:793800. 15. Surh CD, Ahmed-Ansari A, Gershwin ME. Comparative epitope mapping of murine monoclonal and human autoantibodies to human PDH-E2, the major mitochondrial autoantigen of primary biliary cirrhosis. J Immunol 1990;144:2647-2652. 16. Robertson CA, Coppel RL, Prindiville T, Fregeau D, Kaplan M, Dickson ER, Gershwin ME. The relative affinity of recombinant dihydrolipoamide transacetylase for autoantibodies in primary biliary cirrhosis. HEPATOLOGY 1990;11:717-722. 17. Humphries KM, Szweda LI. Selective inactivation of alpha-ketoglutarate dehydrogenase and pyruvate dehydrogenase: reaction of lipoic acid with 4-hydroxy-2-nonenal. Biochemistry 1998;37:15835-15841. 18. Macdonald RA, Hosking CS, Jones CL. The measurement of relative antibody affinity by ELISA using thiocyanate elution. J Immunol Meth 1988;106:191-194. 19. Pullen GR, Fitzgerald MG, Hosking CS. Antibody avidity determination by ELISA using thiocyanate elution. J Immunol Meth 1986;86:83-87. 20. Lam KS, Lake D, Salmon SE, Smith J, Chen ML, Wade S, Abdul-Latif F, et al. A one-bead one-peptide combinatorial library method for B-cell epitope mapping. Methods 1996;9:482-493. 21. Fontenot JD, Ball JM, Miller MA, David CM, Montelaro RC. A survey of potential problems and quality control in peptide synthesis by the fluorenylmethoxycarbonyl procedure. Peptide Res 1991;4:19-25. 22. Rovero P, Quartara L, Fabbri G. Solid-phase synthesis of neurokinin A antagonists. Comparison of the Boc and Fmoc methods. Int J Peptide Protein Res 1991;37:140-144. 23. Howard MJ, Fuller C, Broadhurst RW, Perham RN, Tang JG, Quinn J, Diamond AG, et al. Three-dimensional structure of the major autoantigen in primary biliary cirrhosis. Gastroenterology 1998;115:139-146. 24. Bassendine MF, Jones DE, Yeaman SJ. Biochemistry and autoimmune response to the 2-oxoacid dehydrogenase complexes in primary biliary cirrhosis. Semin Liver Dis 1997;17:49-60. 25. Cha S, Leung PS, Coppel RL, Van de Water J, Ansari AA, Gershwin ME. Heterogeneity of combinatorial human autoantibodies against PDC-E2 and biliary epithelial cells in patients with primary biliary cirrhosis. HEPATOLOGY 1994;20:574-583. 26. Leung PS, Iwayama T, Coppel RL, Gershwin ME. Site-directed mutagenesis of lysine within the immunodominant autoepitope of PDC-E2. HEPATOLOGY 1990;12:1321-1328. 27. Leung PS, Krams S, Munoz S, Surh CP, Ansari A, Kenny T, Robbins DL, et al. Characterization and epitope mapping of human monoclonal antibodies to PDC-E2, the immunodominant autoantigen of primary biliary cirrhosis. J Autoimmun 1992;5:703-718. 28. Moteki S, Leung PS, Dickson ER, Van Thiel DH, Galperin C, Buch T, Alarcon-Segovia D, et al. Epitope mapping and reactivity of autoantibodies to the E2 component of 2-oxoglutarate dehydrogenase complex in primary biliary cirrhosis using recombinant 2-oxoglutarate dehydrogenase complex. HEPATOLOGY 1996;23:436-444. 29. Nishio A, Van de Water J, Leung PS, Joplin R, Neuberger JM, Lake J, Björkland A, et al. Comparative studies of antimitochondrial autoantibodies in sera and bile in primary biliary cirrhosis. HEPATOLOGY 1997;25: 1085-1089. 30. Palmer JM, Diamond AG, Yeaman SJ, Bassendine MF, Jones DE. T cell responses to the putative dominant autoepitope in primary biliary cirrhosis (PBC). Clin Exp Immunol 1999;116:133-139. 31. Rowley MJ, McNeilage LJ, Armstrong JM, Mackay IR. Inhibitory autoantibody to a conformational epitope of the pyruvate dehydrogenase complex, the major autoantigen in primary biliary cirrhosis. Clin Immunol Immunopathol 1991;60:356-370. 32. Rowley MJ, Scealy M, Whisstock JC, Jois JA, Wijeyewickrema LC, Mackay IR. Prediction of the immunodominant epitope of the pyruvate dehydrogenase complex E2 in primary biliary cirrhosis using phage display. J Immunol 2000;164:3413-3419. 33. Yip TT, Van de Water J, Gershwin ME, Coppel RL, Hutchens TW. Cryptic antigenic determinants on the extracellular pyruvate dehydrogenase complex/mimeotope found in primary biliary cirrhosis. A probe by affinity mass spectrometry. J Biol Chem 1996;271:32825-32833. 34. Quinn J, Diamond AG, Palmer JM, Bassendine MF, James OF, Yeaman SJ. Lipoylated and unlipoylated domains of human PDC-E2 as autoantigens in primary biliary cirrhosis: significance of lipoate attachment. HEPATOLOGY 1993;18:1384-1391. 35. Sasaki M, Ansari A, Pumford N, Van de Water J, Leung PSC, Humphries KM, Szweda LI, et al. Comparative immunoreactivity of anti-trifluoroacetyl (TFA) antibody and anti-lipoic acid antibody in primary biliary cirrhosis: searching for a mimic. J Autoimmun 2000;15:51-60. 36. Van de Water J, Fregeau D, Davis P, Ansari A, Danner D, Leung P, Coppel R, et al. Autoantibodies of primary biliary cirrhosis recognize dihydrolipoamide acetyltransferase and inhibit enzyme function. J Immunol 1988; 141:2321-2324. 37. Courvalin JC, Worman HJ. Nuclear envelope protein autoantibodies in primary biliary cirrhosis. Semin Liver Dis 1997;17:79-90. 38. Banga JP, Harris PE. Potential pathogenicity of autoantibodies to thyrotropin receptor in treated, euthyroid patients with Graves’ disease. Eur J Endocrinol 1998;139:139-142. 39. Borza DB, Netzer KO, Leinonen A, Todd P, Cervera J, Saus J, Hudson BG. The goodpasture autoantigen. Identification of multiple cryptic epitopes on the NC1 domain of the alpha3(IV) collagen chain. J Biol Chem 2000; 275:6030-6037. 40. Huang SC, Scofield RH, Kurien BT, Harley JB. Human anti-Ro autoantibodies bind multiple conformational epitopes of 60-kD Ro autoantigen. J Clin Immunol 1997;17:212-219. 41. Kalluri R, Sun MJ, Hudson BG, Neilson EG. The Goodpasture autoantigen. Structural delineation of two immunologically privileged epitopes on alpha3(IV) chain of type IV collagen. J Biol Chem 1996;271:90629068. HEPATOLOGY Vol. 33, No. 4, 2001 42. Kohn LD, Shimura H, Shimura Y, Hidaka A, Giuliani C, Napolitano G, Ohmori M, et al. The thyrotropin receptor. Vitam Horm 1995;50:287384. 43. Brown WR, Kloppel TM. The liver and IgA: immunological, cell biological and clinical implications. HEPATOLOGY 1989;9:763-784. 44. Kaetzel CS, Robinson JK, Chintalacharuvu KR, Vaerman JP, Lamm ME. The polymeric immunoglobulin receptor (secretory component) mediates transport of immune complexes across epithelial cells: a local defense function for IgA. Proc Natl Acad Sci U S A 1991;88:8796-8800. 45. Lamm ME, Robinson JK, Rao CK, Vaerman JP, Kaetzel CS. Epithelial transport of IgA immune complexes. Adv Exp Med Biol 1991;310:187191. 46. Mazanec MB, Lamm ME, Lyn D, Portner A, Nedrud JG. Comparison of IgA versus IgG monoclonal antibodies for passive immunization of the murine respiratory tract. Virus Res 1992;23:1-12. 47. Malmborg AC, Shultz DB, Luton F, Mostov KE, Richly E, Leung PSC, Benson GD, et al. Penetration and co-localization in MDCK cell mito- MIGLIACCIO ET AL. 48. 49. 50. 51. 52. 53. 801 chondria of IgA derived from patients with primary biliary cirrhosis. J Autoimmun 1998;11:573-580. Corthesy B. Recombinant secretory IgA for immune intervention against mucosal pathogens. Biochem Soc Trans 1997;25:471-475. Auer J, Senechal H, De Almeida M. Sperm-associated and circulating IgA and IgG classes of antibodies recognise different antigens on the human sperm plasma membrane. J Reprod Immunol 1997;34:121-136. Russell MW, Sibley DA, Nikolova EB, Tomana M, Mestecky J. IgA antibody as a non-inflammatory regulator of immunity. Biochem Soc Trans 1997;25:466-470. Hexham JM, Carayannopoulos L, Capra JD. Structure and function in IgA. Chem Immunol 1997;65:73-87. Zhang W, Lucisano YM, Lachmann PJ. Complement in IgA immunecomplex-induced neutrophil activation. Biochem Soc Trans 1997;25: 462-466. Sumiyoshi K, Andoh A, Fujiyama Y, Sakumoto H, Bamba T. Characterization of complement C3, C4, and factor B molecules in human bile. J Gastroenterol 1997;32:230-235.